Sample surface preparation - its effect upon the analysis of hydrogen samples

TWI Bulletin, January 1987

by Don Parker and Norman Jenkins

Norman Jenkins, AMet, is Head of the Chemical Laboratory, and Don Parker is a Senior Research Chemist, both in the Materials Department.

The method used for cleaning the sample surface when analysing for diffusible, residual or total hydrogen content can affect the results obtained. Different cleaning methods are examined here, and recommendations made as to the mostappropriate method to use.

A collaborative study to examine the standard ISO method for the determination of diffusible hydrogen in ferritic arc weld metal ^[1] noted that the accuracy and reproducibility of the results depended, in part, upon the surface condition of the sample. It was found that the release rate of diffusible hydrogen was reduced by the presence of an oxide film, but given sufficient time for complete evolution, then the results for oxidised, grit blasted, and wire brushed samples were the same after 21 days' evolution ( Table 1). This effect, which had been reported previously by Boniszewski and Morris, ^[2] was a prime factor in the development of a new standard by the International Institute of Welding, ^[3] which overcame sample surface effects by specifying that the hydrogen should be measured only when evolution had ceased.

Table l Sample cleaning experiment, results for diffusible hydrogen at three days and 21 days (Ref. ^[1] .

The current British Standard 6693: Part 2 ^[4] incorporates this longer collection time, which can be up to 14 days on a cleaned specimen, and describes the preparation of manual metal arc weld deposits to be analysed for diffusible hydrogen by collection over mercury atroom temperature. To reduce the collection time, attempts are being made to replace the mercury method with more rapid methods which involve heating the sample. However, the surface condition of the sample is equally important at the higher temperatures involved, which are variously between l50-650°C, and many workers have studied this aspect of hydrogen analysis.

A report by Laycock ^[5] states that any oxide film left on the surface of the sample leads to significant errors in the results when analysis is carried out at elevated temperatures. This finding was confirmed and reported in ref ^[1] ( Table 2). Similar work by Jenkins ^[6] on the analysis of high carbon steel strip for hydrogen illustrates the possible losses or gains in hydrogen results depending upon the method of cleaning. Cooke and Shanahan ^[7] reported that abrasion of a mild steel surface by silicon carbide paper introduced 0.0049ml H ₂ per square cm into the surface. Further work by Jenkins and Greenfield ^[8] on the analysis of hydrogen standards with different surface conditions showed that either losses or gains of hydrogen could occur, depending upon the sample surface treatment.

Table 2 Sample cleaning experiment results for total hydrogen measured by extraction at 650°C

During the collaborative study referred to above, ^[1] the effect of the sample cleaning method upon residual hydrogen results, measured in vacuum by extraction at 650°C (VHE), was investigated and it was found that grit blasting introduced a large, but variable 'blank'* which was equivalent in the highest instance to as much as 1ml H ₂ /100g of weld on a 4g weld deposit weight. This 'apparent' hydrogen content constitutes a blank which was greater in value than the actual residual hydrogen concentration and, in the case of total hydrogen measured directly by VHE, would form a significant part of the apparent weld hydrogen, ( Table 3).

*'Blank' refers to hydrogen contamination of the surface.

Table 3 Comparison of residual hydrogen measurements on lest welds with different surfaces: a) without further cleaning; b) after recleaning by grit blasting

This article describes further work carried out at The Welding Institute to confirm this finding, to attempt to explain the occurrence of the blank, and seek a means of reducing its value.

Objectives

1. To establish the effect of the sample surface cleaning method on the determination of diffusible, residual and total hydrogen in weld samples made by the standard procedure. ^[4]
2. By comparing different cleaning methods, to select the optimum method in terms of low hydrogen pick-up on the sample surface.
3. To investigate the quantitative effect of analysis temperature upon the measurement of hydrogen introduced into the sample surface by the cleaning operation.

Methods for cleaning Hydrogen analysis samples

Physical cleaning

The cleaning requirement before analysis for hydrogen is removal of oxide film and other contaminants to expose all surfaces of the sample as bright, oxide free metal. Contaminants can range from very hard and closely adhering, such as oxide or slag on a weld surface, to soft, such as paint, grease or general dirt, as on other types of sample. With oxide film or slag, as found on a weld sample, a physically abrasive method must be used, but the inadvertent ablative removal of metal must be kept to a minimum because weighing to determine the weight of weld metal can only be carried out after the hydrogen analysis.

The surface oxide which is formed during welding, and which must be removed before analysis, will depend, in part, upon the squareness of the test block upon which the weld is made, as a poor fit in the welding jig will lead to airgaps and oxidation. Similarly, a rough surface finish could also create an air gap with oxidation potential. However, both these influences would be largely overcome by the use of a physical cleaning method to remove oxide film.

Another factor involved in the choice of cleaning method is that to prevent loss of hydrogen from the sample at room temperature it is stored, before analysis, in either solid carbon dioxide (-80°C) or in liquid nitrogen(-196°C). The sample must also be kept cold during physical cleaning, and therefore the method used must be capable of quickly removing a layer of ice from the sample.

The physical cleaning method used should be suitable for irregular surfaces as found on a weld sample, and shot/grit blasting or wire brushing are acceptable alternatives.

Shot/grit blasting

For the work described in this article, a compressed air supply at 5-7bar was used to raise the abrasive from a hopper and project it through a gun jet against the sample to be cleaned. Surface oxide and all other contaminants are physically removed by impingement of the shot/grit and the surface is cleaned, but left in a distorted state (see Fig). This cleaned surface is contaminated with hydrogenous material which could originate from water, hydrocarbons in the compressed air, or the ablative material itself. With the correct parameters of shot/grit type and size, air pressure and gun jet the method is rapid and independent of sample shape or irregularities, and will deal with ice coated samples down to liquid nitrogen temperatures.

The system may be used with a range of abrasives such as sand, alumina or other types of ceramic material; alternatively glass beads or metal shot of varying sizes may be used. Each of these materials will give different cleaning properties and may result in different amounts of hydrogen being introduced into the sample surface.

Wire brushing

Wire brushing can be either manual or mechanical, but it is less effective than shot/grit blasting because it depends upon the 'energy' of the operator, and the wire bristles rapidly lose their cutting action, after which the surface of the sample becomes polished rather than cleaned. This latter effect makes it particularly difficult to be sure that the oxide film has been removed. An additional difficulty is created by the slowness of the operation and it is usual to re-quench the sample during wire brushing in order to prevent it from reaching room temperature.

Solvent washing

The final sample treatment before any hydrogen analysis is a wash with a suitable organic solvent to remove loose surface debris left by physical cleaning, and frost or moisture which is on the sample because of its low temperature. Thus the solvent must be miscible with water, and have a low boiling point so that it can be dried off the sample in warm air.

It is important that all traces of moisture and solvent are removed from the sample before analysis because they are a source of hydrogen and can lead to a positive error in the results.

Experimental details

Material

The material used for test piece assemblies was a low carbon mild steel. The sample block analysis sections were 30 x 15 x 10mm and were degreased by boiling first in petroleum spirit, GPR ^† (BP range 80-120°C), then in acetone, GPR. Following degreasing they were de-gassed by heating in vacuum for 2hr at 750°C, cooled overnight in the vacuum and stored in a desiccator until required.

^† General purpose reagent

Surface cleaning

To examine the effect of different cleaning methods upon surface contamination by hydrogen, the following techniques were compared.

1. Shot/grit blasting

Several different types of abrasive were used as follows: - Grit, mesh size 24/30;
- Grit, mesh size 80/120;
- Fine mesh metal shot;
- Fine mesh glass beads.

2. Filing

A smooth file was used to remove a small amount of the surface from an unused sample block. This is not a normal cleaning method and could not be used on the irregular surface of a weld, but it has been included in these tests for comparison.

3. Wire brushing

A new, five row, steel wire brush was used manually.

Solvent washing

Every sample was washed before analysis in a stream of acetone to remove debris and water film, which was then removed by evaporation in warm air from a blower.

Welding

All weld samples were made using a 4mm diameter E8018 manual metal arc (MMA) electrode, following procedures given in ISO 3690. The welding parameters were electrode positive at 185A. These conditions gave a deposited weld metal weight of approximately 4g.

Analysis for hydrogen

Two types of hydrogen measurement were carried out and all hydrogen concentrations were calculated in the deposited weight of weld metal.

Diffusible hydrogen

The room temperature diffusible hydrogen was measured by collection over mercury in a Y tube type eudiometer. The technique is fully described in the International Institute of Welding Draft Standard and follows the same principle as the ISO Standard.

Residual and total hydrogen

The determination of residual and total hydrogen was by hot extraction from the sample in a vacuum furnace, followed by measurement of the partial pressure of hydrogen in the vacuum system.

Table 4 Hydrogen content of test blocks: a) before de-gassing; b) after de-gassing and grit blasting

Details of tests and results

To achieve the objectives given earlier the following series of tests were carried out.

1 Hydrogen tests on the test block material and its surface

Six as-ground test sample blocks, size 30 x 15 x 10mm, were degreased as described above and analysed for total hydrogen by VHE at 750°C, this being the normal degassing temperature and 100°C above the normal residual hydrogen analysis temperature. As shown in Table 4, there was 0.21 ml H ₂ /100g when calculated on a 4g weld deposit weight.

The same size test pieces were then oxidised slightly at 500°C to produce a surface oxide similar to that occurring in welding, quenched in cold water and solid CO ₂ . The samples were brought up to room temperature in an acetone rinse, dried in warm air, and placed under mercury in a Y tube for 21 days. No diffusible hydrogen was introduced by quenching from 500°C, or by the rinsing and drying procedure.

After removal from the Y tube, the same samples were loaded overnight in the vacuum system and then analysed for hydrogen by VHE at 650°C. Small amounts of surface moisture or solvent are removed by this overnight storage. The hydrogen pick-up on the surface caused by the cleaning was 0.42 ml/100g when calculated on a 4g weld deposit weight.

2 Surface blank from different cleaning methods

The range of cleaning abrasives and methods described above were applied to standard size sample blocks which had been vacuum degassed, but not given any subsequent oxidation treatment. During cleaning the samples were kept cold and, if necessary; were re-quenched in a solid CO ₂ /industrial methylated spirit (IMS) slurry.

After cleaning the samples were raised to room temperature in an acetone rinse, dried in warm air and analysed for hydrogen by VHE at 650°C. The results are given in Table 5a calculated on a 4g weld deposit weight.

The time taken to carry out the different cleaning procedures was determined on complete welded test piece assemblies, only the centre section being cleaned. The times for complete removal of oxide film are given in Table 5b.

Table 5 a) Sample blocks analysed for hydrogen after different surface treatments

Table 5 b) Cleaning times for different surface treatments on weld samples

3 The effect of further cleaning after oxide removal

Further tests were carried out on the same de-gassed samples as in Table 5, to determine whether the apparent surface hydrogen introduced by the cleaning method could be reduced by a subsequent, more intensive, physical cleaning operation. Three cleaning levels were investigated, each being carried out on samples cleaned by the methods given in Table 6, followed by quenching in CO ₂ /IMS:

Level 1 The samples were wiped with a paper tissue;
Level 2 The samples were agitated in acetone for 30sec in an ultrasonic cleaning bath;
Level 3 The samples were wire brushed, then agitated in acetone for 3osec in an ultrasonic cleaning bath.

After the additional cleaning treatment, the samples were rinsed with acetone, dried in warm air, and loaded into the vacuum system to be stored overnight before analysis. For each cleaning operation four samples were treated and analysed by VHE at 650°C. The mean of each measurement is given in Table 6, calculated on a 4g weld deposit weight.

Table 6 The effect of further cleaning treatment on weld samples cleaned initially by methods given in Table 5

4 Influence of analysis temperature on surface hydrogen measurement

The extraction temperature normally used for measuring total hydrogen in ferritic materials is 650°C. However, the surface blank introduced by physical cleaning has uncertain origins and it was thought necessary to determine the apparent hydrogen concentration measured at different temperatures.

Tests were carried out on samples which had been de-gassed at 750°C for 2hr, stored in CO ₂ quenched and then cleaned by grit blasting with 24/30 mesh grit, followed by cleaning to level 3, as described above, rinsing with acetone, drying in warm air and loading at room temperature in the vacuum system overnight. A liquid nitrogen cold trap was in place and, as with the Y tubes, no diffusible hydrogen was detected. This overnight vacuum storage also had the effect of removing volatile surface contaminants which could be a source of hydrogen. A range of analysis temperatures from 50 to 950°C was used, the samples being analysed by VHE for 1hr at each temperature. The samples initially analysed at temperatures below 650°C were given a further 1hr extraction at 650°C whilst samples analysed initially at 650°C and above were given a further analysis at 950°C.

For each analysis temperature, four samples were analysed and the means of the results are reported in Table 7, calculated on a 4g weld deposit weight.

Table 7 Hydrogen blanks determined at different temperatures on grit blasted sample blocks (24/30 mesh)

The effect of different surface cleaning methods upon manual metal arc weld hydrogen

Twenty MMA welds were made following the ISO 3690 procedure. The five cleaning methods listed in Table 8 were applied to batches of four welds followed by a further cleaning to level 3. After the determination of diffusible hydrogen, with a collection time of 35 days to ensure complete evolution of hydrogen, the samples were transferred from the Y tubes to the VHE equipment for the measurement of residual hydrogen at 650°C. The residual hydrogens were corrected for the appropriate surface treatment hydrogen values. The means of the results are reported in Table 8.

Table 8 Effect of different surface cleaning methods upon residual and total hydrogen content of standard test welds

Discussion

Analysis results

The results for diffusible hydrogen which are given in Table 1 show the marked effect of sample surface condition upon the evolution rate of hydrogen from a standard weld sample. It is clear that the non removal, or inefficient removal, of surface oxide delays the release of hydrogen, but this effect may be countered by allowing sufficient time for complete evolution before measuring the hydrogen. However, Table 2 demonstrates that when determining total hydrogen, as for example by vacuum hot extraction, then the presence of a surface oxide causes some of the evolved hydrogen to be oxidised to water, leading to a very low result, i.e. 28% of the hydrogen was lost. Thus, if the sample is to be heated to recover total hydrogen directly, or to measure residual hydrogen after the measurement of diffusible hydrogen, then some form of cleaning the sample to remove surface oxide must be employed. The effects of three types of surface cleaning method upon residual hydrogen measurements are given in Table 3. It is seen that grit blasting has produced a higher apparent residual hydrogen content. Further tests in which half the samples in Table 3 were given a further grit blasting treatment showed that this cleaning method introduced a surface contamination which released hydrogen from the sample only when it was heated for hydrogen analysis. No diffusible hydrogen was released from these grit blasted samples during overnight storage in a vacuum, but on heating for analysis at 650°C about 0.5 ml/100g hydrogen was measured and at this temperature a positive hydrogen blank is introduced by grit blasting, but the samples given a second grit blasting treatment did not have their apparent hydrogen 'blank' further increased. However, in making comparisons between the residual hydrogen results in Table 3, the 0.1 ml/1 00g residual for the 'as welded' will be low because of the presence of a partial oxide film. Therefore, the true residual for these eight welds will be between 0.1 ml/100g and 0.5 ml/100g.

This finding was investigated further by tests on the sample test blocks ( Table 4). Six test blocks, 30 x 15 x 10mm, were degreased and analysed in the as-ground condition for hydrogen by VHE at 750°C. The mean hydrogen in the test block, before degassing, was only 0.0082ml at STP, but this represents 0.21 ml/100g on a 4g weld deposit weight. After analysis, the six test blocks were oxidised, water quenched, and then grit blasted and re-analysed by VHE at 650°C, and it was found that the mean hydrogen concentration introduced by the grit blasting was 0.42 ml/100g. These results indicate the importance of using de-gassed steel test blocks for the standard welds if subsequent analysis is to be by hot extraction, and confirm the blank hydrogen introduced by grit blasting. It is possible that part of the 0.21 ml/100g of hydrogen found during the analysis at 750°C of the as-ground test blocks is also a grinding blank and may be subject to the same variability as the apparent hydrogen blank found by grit blasting. ^[7]

The origin of the blank hydrogen is uncertain, but it is probable that moisture on the sample, rubber dust from the flexible couplings, or other types of hydrocarbon from the grit blast machine are physically trapped in the sample surface by the ablative/ compressive action of the grit. If this were the case, then different types of cleaning material might be expected to differ in their effect on the sample surface through either different densities, hardness, or particle size and surface geometry.

A range of surface cleaning techniques was examined and the results in Table 5 indicate that different amounts of blank hydrogen are introduced by different 'grits' or cleaning methods. The 24/30 mesh grit blasting is again confirmed as a source of about 0.5 ml/100g hydrogen, as in Table 3, but the smaller grit size, 80/120 mesh, of the same material and the fine glass beads introduced significantly less hydrogen into the sample block surface. The lowest hydrogen blank was obtained when using fine metal beads, but the cleaning process was much slower with this abrasive and the equipment used ( Table 5b). Filing the cold, wet surface was successful in avoiding hydrogen pick-up by the sample block surface, but this method of cleaning was only included for comparison. Filing is of little practical use in preparing weld samples for analysis. It was found that vigorous wire brushing introduced the lowest hydrogen blank, equivalent to 0.03 ml/100g when calculated on a 4g weld deposit weight. The results for the total hydrogen measurement on wire brushed samples ( Table 2) indicate that wire brushing is capable of removing the oxide film from the weld surface, but it is emphasised that manual wire brushing involves a subjective assessment of the degree of oxide removal. Further, wire brushing is less rapid than grit blasting, and a much used wire brush is less efficient at removing oxide than a new brush. Thus, for best results, brushes should be discarded frequently, possibly after only 6-8 samples.

The apparent hydrogen blank obtained on the wire brushed sample was 0.03 ml/100g, and did not merit further tests. However, it was not possible to evaluate the degree of oxide removal by wire brushing and it should be noted that even a small amount of oxide film on the surface will reduce the hydrogen recovery. However, the blank from grit and shot blasting was sufficient to have a significant effect on both total and residual hydrogen measurements. The effect of further surface treatment using three levels of cleaning was investigated ( Table 6). It was found that ultrasonic cleaning in a bath of acetone visibly removed small amounts of debris from the surface of the sample, but the reduction in hydrogen was minimal, and in one instance (grit blasting, 80/120 mesh) there was an increase in hydrogen. Even when cleaned to level 3 (wire brushing followed by ultrasonic cleaning) the apparent hydrogen was little changed, although the removed debris was magnetic, showing that the wire brushing had removed some metal. It was evident from these tests that the surface contamination was physically very tightly trapped in the sample surface and that it could not be removed completely by a subsequent cleaning treatment.

The blank hydrogen concentrations reported in Tables 5 and 6 were measured by VHE at 650°C for 1 hr, and further measurements on de-gassed grit blasted sample blocks (24/30 mesh) were carried out over a range of extraction temperatures to establish the temperature dependence of the blank. The results ( Table 7) indicate a strong temperature dependence, the blank increasing markedly above 250°C, and again at 650°C and above. When the same sample blocks were subjected to a second analysis at 650°C or 950°C, as appropriate, further hydrogen was extracted. When the total hydrogen blank is considered, up to 650°C, the amount measured is comparable with the results reported in Tables 3-6 but at 950°C a substantial increase in the blank takes place. This increase in blank with increased temperature is compatible with the source of hydrogen being some form of contamination, such as grit, rubber dust, oil or water, sealed into the surface and only dissociating to hydrogen at the higher temperatures.

Previous unpublished work showed that the use of compressed air from a cylinder did not reduce the surface hydrogen pick-up and surface analysis by X-ray photoelectron spectroscopy showed increased concentrations of oxygen and carbon. The source of the oxygen was assumed to be the silicate grit which was used and the carbon binding energy was identified as that associated with hydrocarbons.

It is common in many types of measurement to encounter a blank which contributes a positive (or negative) error to the quantity being evaluated. Provided the blank value can be accurately assessed, then true values for the measurand are arrived at by subtracting the blank from the measured quantity. The quantitative effect of the apparent hydrogen blank caused by different cleaning techniques was investigated in a series of weld hydrogen measurements. Diffusible and residual hydrogen were measured and the blank values reported in Table 6 were subtracted from the results obtained ( Table 8).

Implications

Differences between cleaning methods are small in relation to the total hydrogen, in this case about 9.2 ml/100g of weld deposit, and the errors introduced by not making the blank correction are not great when compared with the ±10% overall reproducibility of the diffusible hydrogen method. Nevertheless, the overall effect of a hydrogen blank, caused by heating a sample cleaned by grit blasting for analysis, is a consistently positive but variable bias of the hydrogen result. This consistent bias, although small, assumes great importance when a result for diffusible hydrogen is assessed according to the BS 5135 classification, i.e. placed with the 5, 5-10, 10-15, and 15 ml/100g hydrogen bands. ^[8] Further, the proportional error of a 0.4 ml/100g bias is greater at the lower levels of hydrogen which are achieved using current types of low hydrogen electrodes.

The positive bias is minimised by the selection of a cleaning technique which gives a low apparent hydrogen blank, such as wire brushing. However, this method is slower than grit blasting, necessitating re-quenching of the sample, there is difficulty because of the presence of an ice layer, and the extent of oxide removal is a subjective judgement. When using a wire brushing technique it is important to use a new wire brush as failure to remove all of the oxide film will introduce a negative bias.

Summary

1. Grit blasting of weld hydrogen samples introduces a positive blank apparent hydrogen into the sample surface which does not affect the room temperature diffusible hydrogen, but is released when analysis takes place by a method involving heating above about 250°C.
2. It is probable that distortion of the sample surface during grit blasting causes physical trapping of some form of contaminant (grit, rubber, dust, oil, water) within the surface. This encapsulated source of hydrogen undergoes increasing dissociation as the temperature of analysis is increased to 950°C.
3. Shot blasting, with metal or glass, results in a lower hydrogen blank than obtained by grit blasting, but metal shot blasting is slow with the equipment used.
4. Wire brushing is slower than grit blasting, but appears to be capable of cleaning the sample of oxide film and results in the lowest apparent hydrogen blank.

Recommendations

Grit or shot blasting removes oxide film and is recommended for use on diffusible hydrogen samples which are to be analysed between room temperature and 250°C. Grit or shot blasting should not be used when the samples are to be analysed at temperatures above 250°C.

On the basis that grit or shot blasting is not an acceptable cleaning method for analysis at above 250°C, the only practical alternative is wire brushing. However, new brushes must be used and complete removal of oxide is dependent upon the competence of the operator.

References